Drift into failure, p.23
Drift Into Failure, page 23
One of my students recently argued that we really shouldn't be too pessimistic in our outlook for safety. Aviation, as a world-wide industry, is not drifting into failure. After all, he said, haven't we made enormous progress? Look at the accident statistics and the number of fatalities, he said. Of course, there are spikes and variations. But year over year, we seem to get better and better, not drifting into something worse.
Well, another said, this depends on how you define progress, and where you draw the boundary of the fatalities that the system may be thought to be responsible for (and, indeed, how you define such "responsibility"). We generate energy in aviation the very same way as we did in pre-history. We burn carbon-based material, just as we did when we were dwelling in caves. Doing the same thing for 10,000 years? We should not really count that as progress. What's more, how do we know that this is producing fewer fatalities? Doesn't that depend on where we look for system interconnections and where we start and stop counting dead bodies? The traditional way of assessing fatalities in aviation is to count the victims in the crashed airplane and any additional victims on the ground. That seems to make good sense, of course. But perhaps it is also a quite limited way of looking at the fatal consequences of flying. And, in a sense, a very parochial and self-indulgent one. We count only those in the world who were already fortunate (and affluent) enough to take direct part in the activity (which in itself is an increasingly sizable portion of humanity). But what about other people who are affected?
For example, we might want to look for connections between aviation activity, global warming, and increasing wars in Africa. The body count would go up dramatically, changing the notion of "progress" if only we are willing to include a bigger slice of humanity in the picture. Since 1990, emissions of carbon dioxide by aviation have risen 90 percent. According to some reports, aviation's share of CO2 emissions could rise from about 2 percent globally today, to as much as 20 percent in 2050.18 Of course, enormous economic opportunities are created by such global interconnections and their increase, in ways that are not very dependent on land-based infrastructure. So that could count as progress along a number of dimensions, for sure.
But there are other effects. The release of CO2 into the atmosphere (each kilogram of aviation fuel that is burned basically adds two kilograms of CO2 to the atmosphere) has been linked not only to a global increase in temperature, but to changes in regional climates as well.19 The earth, of course, has always known temperature cycles and fluctuations in atmospheric gas composition. And only some of the CO2 in the atmosphere comes from human activity. Since the dawn of the Industrial Revolution, however, this is an increasing (and, lately, accelerating) proportion. Today the coupling between anthropogenic CO2 and climate change (or global warming) is seen as proven and lawful by many; others dismiss it as amusing and as junk science.
But changes in CO2 concentrations (whatever their source) do have effects on regional climates. Using what they called a zonally symmetric synchronously coupled biosphere-atmosphere model, NASA and MIT scientists showed a connection between CO2 concentrations and changes in the biosphere-atmosphere system in Africa. Changes in CO2 concentration make certain regions wetter,20 and, by extension, certain regions drier.21 Other scientists have found a strong link between drier years and the prevalence of conflict in Africa. Long-term fluctuations of war frequency follow cycles of temperature change: there is more conflict during warmer years. Rain-fed agriculture accounts for more than 50 percent of gross domestic product, and up to 90 percent of employment, across much of the continent. A drier, warmer climate decreases agricultural output (a 10–30 percent drop per °C of warming), which in turn degrades economic performance, which in turn increases conflict incidence. This holds even after correcting for per capita income or democratization (increases in both tend to reduce war risk).
Warming, then, has a role in shaping conflict risk, with projected climate changes increasing armed conflict by 54 percent and bringing an additional 393,000 African deaths by 2030.22 That would be like 20,000 deaths each year. If aviation's share of CO2 production indeed rises from 2 percent to 20 percent over the next 40 years, its share in those deaths would drift from about 400 per year today to about 2,000 fatalities per year in 2030. In aviation, such figures would not count as progress; it could count as another slow, steady drift into failure. Also, it would outpace any beneficial effects of African economic growth and democratization (some of which, paradoxically, are also brought by aviation with its role in stimulating economic development and bringing people and ideas and goods and services together).
Of course, there are a lot of numbers in the example above. But the example is not about the numbers, it is about the possible relationships, and about where we believe we should draw the boundaries and call something a "system." Such boundaries are always exclusionary. Wherever we draw them, something will fall outside of it. Systems thinking is not saying that there should be no boundaries (because they are an essential part of defining what a system is), but that we can at least be more flexible, more imaginary, and perhaps more democratic in where we draw those boundaries.
And, as in all complex systems, other factors could play a plausible role too. All explanations for how system behavior is the result of relationships are tentative, and open for change when better arguments come in. It is not necessary, for example, that additional conflict deaths due to temperature increases are the result of regional warming There is a known correlation between violent crime and higher temperatures, which could account for at least some of those additional deaths,23 and even non-farm labor productivity can decline with higher temperatures, meaning that declines in agricultural output do not have to act as an intermediary variable between more heat and more deaths.24 It is consistent with complexity thinking, then, when researchers say 'We interpret our results as evidence of the strength of the temperature effect rather than as a documentation of the precise future contribution of economic progress or democratization to conflict risk. Similarly, we do not explicitly account for any adaptations that might occur within or outside agriculture that could lessen these countries' sensitivities to high temperatures, and thus our 2030 results should be viewed as projections rather than predictions."25 Being modest about the reach of one's results, and explicitly open to their being wrong or open for revision, was once seen as a weakness in heavily naturalized science. But in complexity and systems thinking, it is a strength.
The example shows how physics was by no means left alone in this drift away from classical Newtonian science. In fact, systems dunking was in part pioneered by biologists, who began to look at living systems not as collections of components but as irreducible, integrated wholes. Social sciences and humanities have similarly put up ever stronger resistance over the past few decades against the wholesale naturalization of their research work. The only way to be "scientific" was once to be like physics (in fact, for centuries being more scientific meant having to be like physics). That meant that research had to be done in the form of carefully controlled experiments, in which the scientist had clear control over one or a few variables, tweaking them so that she or she or he could exactly understand what had caused what. Preferably, the results were expressed as, or converted into, numbers, so that the claim that something fundamental had been found about the world looked even more credible. This is the way of doing science that Newton and Descartes proposed. Ironically, with a shift of natural sciences into systems thinking and Complexity, and social sciences and humanities doing the same, it seems that they once again seem to come closer to each other. Not because the social sciences and humanities are becoming "harder" and more quantifiable, but because natural sciences are becoming "softer" with an emphasis on unpredictability, irreducibility, non-linearity, time-irreversibility, adaptivity, self-organization, emergence – the sort of things that may always have been better suited to capture the social order.
It is interesting to see that complexity and systems thinking predates what we regard as the Scientific Revolution in the sixteenth and seventeenth centuries – a revolution that gave rise to the Newtonian–Cartesian hegemony. Leonardo (born Leonardo di Ser Piero, Vinci, Florence, 1452–1519) is regarded as science's first system thinker and complexity theorist. An artist, engineer, architect, relentless empiricist, experimentalist and inventor, Leonardo embodied the humanist fusion of art and science. He embraced a profound sense of the interrelatedness of things, interconnecting observations and ideas from different disciplines (for example, physiology and mechanical engineering; aerodynamics and music) so as to see problems in a completely new light and come up with nifty solutions. Leonardo embraced his science as complexity – it had no boundaries since his goal was to combine, advance, investigate and understand processes in the natural world through an inter-disciplinarian view. He was a committed systems thinker, eschewing mechanistic explanations of phenomena and instead giving primacy to ecological ones – thinking up and out, rather than only down and in, and always seeing (and, in his art, depicting) linkages among living organisms (people, groups, organizations) so as to reveal new solutions. Leonardo was fascinated by the phenomenon of flight and produced many studies on the flight of birds (for example, his 1505 Codex). He designed a variety of aircraft – of which one, a hang glider, was recently demonstrated to actually work.
Complex Systems Theory
With his formulation of General Systems Theory, Von Bertalanffy helped establish a serious scientific foundation for the alternative to Newtonian–Cartesian thought in the early 1970s.26 Today that alternative is known as complexity and systems theory. Recently, Cilliers summarized the characteristics of complex, as opposed to Newtonian, systems nicely.27 I sum the points up as follows, and will explain them in more detail below:
❍ Complex systems are open systems – open to influences from the environment in which they operate and influencing that environment in return. Such openness means that it is difficult to frame the boundaries around a system of interest.
❍ In a complex system, each component is ignorant of the behavior of the system as a whole, and doesn't know the full effects of its actions either. Components respond locally to information presented by them there and then. Complexity arises from the huge, multiplied webs of relationships and interactions that result from these local actions.
❍ Complexity is a feature of the system, not of components inside it. The knowledge of each component is limited and local, and there is no component that possesses enough capacity to represent the complexity of the entire system in that component itself. This is why the behavior of the system cannot be reduced to the behavior of the constituent components, but only characterized on the basis of the multitude of ever-changing relationships between them.
❍ Complex systems operate under conditions far from equilibrium. Inputs need to be made the whole time by its components in order to keep it functioning. Without that constant flow of actions, of inputs, it cannot survive in a changing environment. The performance of complex systems is typically optimized at the edge of chaos, just before system behavior will become unrecognizably turbulent.
❍ Complex systems have a history, a path-dependence. Their past is co-responsible for their present behavior, and descriptions of complexity have to take history into account.
❍ Interactions in complex systems are non-linear. That means that there is an asymmetry between, for example, input and output, and that small events can produce large results. The existence of feedback loops means that complex systems can contain multipliers (where more of one means more of the other, in turn leading to more of one, and so forth) and butterfly effects.
Complex systems are not closed. They don't act in a vacuum like Newton's planetary system. Socio-technical systems or organizations are open systems. They are in constant interplay with their changing environment, buffeted and influenced by what goes on in there, and influencing it in turn. Whereas the systems studied by high reliability theory (see Chapter 4) were relatively closed (an aircraft carrier at sea, power grids before deregulation, air traffic control run by a single government entity, the Federal Aviation Administration), this is not true for many other safety critical systems. NASA, for example, has operated in a (geo-)political and societal force field for decades, suffused with budgets and operational priorities and expectations that never were her own but seeped in from a number of directions and players (congress, the defense department, other commercial space operators) and affected what were seen as rational tradeoffs between acute production pressures and chronic safety concerns at the time, Of course the influence doesn't just go one way. NASA itself also affected what politicians saw as reasonable and doable, in part because of its own production, and its actions would have affected the actions of other commercial space operators too (for example, the choice by the French Ariane consortium not to use manned vehicles for the launching of satellites). This same openness goes for healthcare too: it often is a handmaiden of local or national politics and funding battles – hay gets made from promises to bring down surgical waiting lists, for instance. This can have effects on how funding is allocated, on where resources are added and where they are taken away.
Open systems mean that it can be quite difficult to define the border of a system. What belongs to the system, and what doesn't? This is known as the frame problem. It is very difficult to explicitly specify which conditions are not affected by an action. When you trace the ever-changing webs of relationships in the mining of Coltan for the production of cell phones, for example (see above), you will find that actions by local people reverberate across an almost infinite range of economies and ecologies, affecting not only wildlife in the eastern Congo, but everything from inflation of food prices in local villages surrounding Kahuzi-Biega, to the allure of discarded Russian cargo planes and those who remember how to fly them, to the volatility of commodity markets and demands of just-in-time production of electronics in Taiwan, to the prosperity of particular factions in the underworld of Belgium. And all of these reverberations produce further reverberations, in Russia, Belgium, Taiwan, affecting people and markets and all kinds of relationships there too.
One solution of complexity and systems theory is to determine the scope of the system by the purpose of the description of the system, not by the system itself. If the purpose is to describe the effects of Coltan mining on local gorilla populations, then you might draw the geographic system boundaries not far from Kahuzi-Biega. Then you can trace ecological contamination and physical habitat destruction, and you might restrict functional boundaries to bush-meat prices, hunting methods, and numbers of miner mouths to feed. Never mind Russia or Belgium or Taiwan. But such boundaries are a choice; a choice that is governed by what you want to find out. System theory itself provides no answer to the frame problem. Where you place the frame is up to you, and up to the question you wish to examine. The adage of forensic science to "follow the money" is the same commitment. On the one hand, it leaves the observer or investigator entirely open to where the trail may take her or him and how it branches out into multiple directions, organizations, countries. That is where the system of interest is open. On the other hand, the commitment frames the system of interest as that which can be expressed monetarily.
This highlights a very important aspect of the post-Newtonian perspective: the world we can observe isn't just there, completely and perfectly ready-formed and waiting for us to discover. We ourselves play a very active role in creating the world we observe, precisely because of our observations. If we don't have the language or the knowledge to see something, we won't ever see it. What our observations come up with is in large part up to us, the observers. In post-Newtonian science, it is very hard to separate the observer from the observed; to say where one ends and the other begins. In post-Newtonian science, reality does not contain the sort of hard, immutable facts that have a life entirely independent of a human observer. Human observation cannot be the neutral arbiter or producer of knowledge. After all, the observers themselves impose a particular language, interest, and they come with imaginations and a background that actively bring certain aspects of a problem into view, while leaving others obscured and unexamined.
In a complex system, each component is ignorant of the behavior of the system as a whole. This is a very important point.28 If each component "knew" what effects its actions had on the entire rest of the system, then all of the system's complexity would have to be present in that component. It isn't. This is the whole point of complexity and systems theory. Single elements do not contain all the Complexity of the system. If they did, then reductionism Could work as an analytic strategy: we could explain the whole simply by looking at the part. But in complex systems, we can't, and analytic reduction doesn't work to enhance anybody's understanding of the system.
Complexity is the result of a rich interaction, of constantly evolving relationships between components and the information and other exchanges that they produce. Rich interaction takes a lot of components, which is indeed what complex systems consist of. Of course a beach has a lot of components (sand kernels) too, but that alone doesn't make it complex. The components have to interact. They have to give each other things like energy, information, goods. Those interactions, however, are limited, or local. Each component responds only to the limited information it is presented with, and presented with locally. The Coltan miner has no idea about the price of retired Russian cargo planes, and might not even recognize one if he saw it. What he mines is taken away through the bush on the backs of other men, perhaps with the use of animals or motorbikes, then onto trucks if they can make it that far into the jungle. He may not even know what he is mining, what the stuff is for, or why it is worth so much. The pilot flying the plane to a nearby airstrip is probably ignorant about gorillas, and may not know who really offloads his cargo at the receiving end. He makes the trip not because he works for a major cell phone manufacturer, but because he gets U.S. dollars in bundles of cash at the end of it.
